Document 10813131

advertisement
Gen. Math. Notes, Vol. 6, No. 2, October 2011, pp.45-79
c
ISSN 2219-7184; Copyright ICSRS
Publication, 2011
www.i-csrs.org
Available free online at http://www.geman.in
Multipoint Algorithms Arising from Optimal
in the Sense of Kung-Traub Iterative
Procedures for Numerical Solution
of Nonlinear Equations
Boryana Ignatova1 , Nikolay Kyurkchiev2 and Anton Iliev3
1,2,3
Faculty of Mathematics and Informatics
Paisii Hilendarski University of Plovdiv
236, Bulgaria Blvd., 4003 Plovdiv, Bulgaria
2,3
Institute of Mathematics and Informatics,
Bulgarian Academy of Sciences
Acad. Georgi Bonchev Str., bl. 8
1113 Sofia, Bulgaria
1
E-mail: gboryanaok@abv.bg
2
E-mail: nkyurk@uni-plovdiv.bg
3
E-mail: aii@uni-plovdiv.bg
(Received: 2-8-11/ Accepted: 14-10-11)
Abstract
In this paper we will examine self-accelerating in terms of convergence speed
and the corresponding index of efficiency in the sense of Ostrowski - Traub
of certain standard and most commonly used in practice multipoint iterative
methods using several initial approximations for numerical solution of nonlinear equations (method regula falsi, modifications of Euler - Chebyshev method,
Halley method, and others) due to optimal in the sense of the Kung-Traub algorithm of order 4 and 8. Some hypothetical iterative procedures generated by
algorithms from order of convergence 16 and 32 are also studied (the receipt
and publication of which is a matter of time, having in mind the increased
interest in such optimal algorithms). The corresponding model theorems for
their convergence speed and efficiency index have been formulated and proved.
46
Boryana Ignatova et al.
Keywords: solving nonlinear equations, order of convergence, optimal
algorithm, efficiency index.
1
Introduction
Traub [25] proposed the concept of efficiency index as a measure for comparing
of different methods for solving nonlinear equation
f (x) = 0.
(1)
This index is described by:
1
τ n,
where τ is the order of convergence and n is the whole number of evaluations
per iteration.
Kung and Traub [10] then presented a hypothesis on optimality of rootsolvers by giving
n−1
2 n
as the optimal order of convergence.
This means that the Newton’s method
xk+1 = xk −
f (xk )
f 0 (xk )
k = 0, 1, 2, . . .
by two evaluations per iteration is optimal with 1.414 as the efficiency index.
By taking into account the optimality concept, many authors have tried to
build iterative procedures of optimal order of convergence - τ = 4, τ = 8.
The recent results of M. Petkovic [15] and M. Petkovic and L. Petkovic [16],
Bi, Wu and Ren [2], Geum and Kim [4], Thurkal and Petkovic [24], Wang and
Liu [26], Kou, Wang and Sun [9], Chun and Neta [3], Khattri and Abbasbandy
[8], Soleymani [20], Bi, Ren and Wu [1] are presented for optimal multipoint
methods for solving nonlinear equations.
M. Petkovic [15] gives a useful detailed review about computational efficiency of many methods in the sense of Kung - Traub hypothesis.
For other nontrivial methods for solving nonlinear equations see, Kyurkchiev
and Iliev [11] and monograph by Iliev and Kyurkchiev [7].
In many natural science tasks, from purely physical considerations, for the
user of numerical algorithms for solving nonlinear equation (1) is preliminary
known system of initial approximations
x01 , x02 , . . . , x0t
47
Multipoint Algorithms Arising from...
for the root ξ of equation (1).
As an example, regula falsi methods and modifications of Euler - Chebyshev
method and Halley method with a lower order of convergence using two or three
initial approximations for the root ξ.
In [13], refined conditions of convergence for the difference analogue of Halley method (using three initial approximations) for solving nonlinear equation
are given (see, also [27]).
An efficient modification of a finite - difference analogue of Halley method
is proposed in [6].
Naturally arises the task of designing and testing multipoint variants of
the classical procedures in the light of the achievements over the past five
years important theoretical results related to obtaining optimal in the sense of
Kung-Traub algorithms.
In this sense the task of detailed refinement of the self-accelerating multipoint methods using several initial approximations is actual.
2
Main Results
In this paragraph we will begin considerations of the important issue of selfaccelerating the most frequently use method it regula falsi with the technique
of input optimal in the sense of Kung-Traub algorithms of order τ = 4.
I. The regula falsi method given by
xn+1 = xn −
f (xn )(xn−1 − xn )
,
f (xn−1 ) − f (xn )
(2)
n = 0, 1, 2, . . .
requires 2 function evaluations, 2 initial approximations x−1 , x0 and has order
of convergence τ ≈ 1.618 and efficiency index
1
I = 1.618 2 ≈ 1.272.
The method given by
f (xn )
,
f 0 (xn )
yn
= xn −
xn+1
f (xn )
f (yn )
.
,
= yn − 0
f (xn ) f (xn ) − 2f (yn )
(3)
n = 0, 1, 2, . . .
or
48
Boryana Ignatova et al.
!
xn+1
f (xn )
f xn − 0
f (xn )
f (xn )
f (xn )
!,
= xn − 0
− 0
.
f (xn ) f (xn )
f (xn )
f (xn ) − 2f xn − 0
f (xn )
(4)
n = 0, 1, 2, . . .
proposed by Ostrowski in [14] was the first multipoint method of the fourth
order.
This method requires 3 functional evaluations and has the efficiency index
1
I = 4 3 ≈ 1.587.
Its order of convergence is optimal in the sense of the Kung-Traub conjec2
ture (efficiency index is I = 2 3 ).
We will explicitly mention that there are other algorithms with optimal
order of convergence τ = 4. We will explore Ostrowski iteration as the most
popular representative of this class of optimal methods.
Here we give a methodological construction of nonstationary algorithms
with a raised speed of convergence.
I.1. Let us consider the following nonstationary iterative scheme based on
schemes (2) and (4):
x2n+1 = x2n −
f (x2n )(x2n − x2n−1 )
,
f (x2n ) − f (x2n−1 )
!
x2n+2
f (x2n+1 )
f x2n+1 − 0
f (x2n+1 )
f (x2n+1 )
f (x2n+1 )
!,
= x2n+1 − 0
− 0
.
f (x2n+1 ) f (x2n+1 )
f (x2n+1 )
f (x2n+1 ) − 2f x2n+1 − 0
f (x2n+1 )
(5)
n = 0, 1, 2, . . .
Let
ei = xi − ξ, i = −1, 0, 1, . . . ; ci (ξ) =
f (i) (ξ)
, i = 2, 3, . . . ,
f 0 (ξ)i!
It is well-known that for the error i [25] is valid
49
Multipoint Algorithms Arising from...
2n+1 ∼ −c2 (ξ)2n 2n−1 ,
(6)
and for the procedure (4) [25]:
2n+2 ∼ c2 (ξ)[c22 (ξ) − c3 (ξ)]e42n+1 ,
(7)
where ∼ denotes the asymptotical equation when n → ∞.
Let
K
= max {|c2 (ξ))|, |c2 (ξ)[c22 (ξ) − c3 (ξ)]|} ,
1
d2n−1 = K 2 |2n−1 |,
d2n
= K|2n |
and let d > 0, and x−1 and x0 be chosen so that the following inequalities
1
d−1 = K 2 |x−1 − ξ| ≤ d < 1,
d0
= K|x0 − ξ| ≤ d < 1
hold true.
From (6) and (7), we have
1
1
d2n+1 = K 2 |2n+1 | ≤ K 2 K|2n ||2n−1 | = d2n d2n−1 ,
(8)
d2n+2 = K|2n+2 | ≤
KK42n+1
= K
1
2
4
42n+1
=
d42n+1 .
Our results concerning the order of convergence generated by (5) are summarized in the following theorem.
Theorem 2.1 Assume that the initial approximations x0 , x−1 are chosen
so that d−1 ≤ d < 1 and d0 ≤ d < 1.
∞
Then for the error of the sequences {x2n }∞
n=0 and {x2n−1 }n=0 determined
by (5), we have
d2n−1 ≤ d2.5
n−1
,
(9)
d2n
≤ d
8.5n−1
,
n = 1, 2, . . .
and the order of convergence of the iteration (5) is τ = 5.
50
Boryana Ignatova et al.
Proof. The proof is by induction with respect to the iteration number n.
For n = 0, from (9), we find
d1 ≤ d.d = d2 ,
d2 ≤ d41 = (d2 )4 = d8 .
For n = 1, we have
d3 ≤ d10 ,
d4 ≤ d40
and (9) is fulfilled.
Let (9) be fulfilled for n ≤ m.
For n = m + 1, from (8) and (9), we have
m−1
d2(m+1)−1 = d2m+1 ≤ d2m d2m−1 ≤ d8.5
.d2.5
m−1
m
d2(m+1) = d2m+2 ≤ d42m+1 < d2.5
4
m−1
= d10.5
m
= d2.5 ,
m
= d8.5
which completes the induction.
On the other hand,
1
d2n−1 = K 2 |2n−1 |,
d2n
= K|2n |
and equation (9) can be written as
1
n−1
|2n−1 | ≤ K − 2 d2.5
|2n |
n−1
≤ K −1 d8.5
,
, n = 1, 2, . . . ,
and the order of convergence of iteration (5) is equal to 5.
Thus, the theorem is proved.
Remark 1. The new method (5) requires 5 function evaluations, 2 initial
approximations x−1 , x0 and has order of convergence τ = 5 and efficiency index
1
I = 5 5 ≈ 1.3797.
Sharma and Sharma [18] presented the following family of optimal order
51
Multipoint Algorithms Arising from...
eight
yn
= xn −
f (xn )
,
f 0 (xn )
zn
= yn −
f (xn )
f (yn )
.
,
0
f (xn ) f (xn ) − 2f (yn )

xn+1
f (zn )
f (zn )
= zn − 1 +
+
f (xn )
f (xn
(10)
!2 
.
f [xn , yn ]f (zn )
.
f [xn , zn ]f [yn , zn ]
n = 0, 1, 2, . . .
Here, f [x, y] denote the finite difference.
This method requires 4 functional evaluations and has the efficiency index
1
I = 8 4 ≈ 1.6817.
Its order of convergence is optimal in the sense of the Kung-Traub conjec3
ture (I = 2 4 ).
I.2. Let us consider the following nonstationary iterative scheme based on
schemes (2) and (10):
x2n+1 = x2n −
f (x2n )(x2n − x2n−1 )
,
f (x2n ) − f (x2n−1 )
y2n+1 = x2n+1 −
f (x2n+1 )
,
f 0 (x2n+1 )
z2n+1 = y2n+1 −
f (y2n+1 )
f (x2n+1 )
.
,
f 0 (x2n+1 ) f (x2n+1 ) − 2f (y2n+1 )

x2n+2
f (z2n+1 )
f (z2n+1 )
= z2n+1 − 1 +
+
f (x2n+1 )
f (x2n+1
!2 
.
f [x2n+1 , y2n+1 ]f (z2n+1 )
,
f [x2n+1 , z2n+1 ]f [y2n+1 , z2n+1 ]
(11)
n = 0, 1, 2, . . .
It is known that for the error 2n+2 = x2n+2 − ξ [18] is valid
2n+2 ∼ A(ξ)e82n+1 ,
(12)
We will use again the fact that for regula falsi method is satisfied
2n+1 ∼ −c2 (ξ)2n 2n−1 .
(13)
52
Boryana Ignatova et al.
Let
K1
= max {|c2 (ξ))|, |A(ξ)|} ,
1
d2n−1 = K14 |2n−1 |,
d2n
= K1 |2n |
and let d > 0, and x−1 and x0 be chosen so that the following inequalities
1
d−1 = K14 |x−1 − ξ| ≤ d < 1,
d0
= K1 |x0 − ξ| ≤ d < 1
hold true.
From (12) and (13), we have
1
1
d2n+1 = K14 |2n+1 | ≤ K14 K1 |2n ||2n−1 | = d2n d2n−1 ,
1
4
d2n+2 = K1 |2n+2 | ≤ K1 K1 82n+1 = K1
8
(14)
82n+1 = d82n+1 .
Our results concerning the order of convergence generated by (11) are summarized in the following theorem.
Theorem 2.2 Assume that the initial approximations x0 , x−1 are chosen
so that d−1 ≤ d < 1 and d0 ≤ d < 1.
∞
Then for the error of the sequences {x2n }∞
n=0 and {x2n−1 }n=0 determined
by (11), we have
d2n−1 ≤ d2.9
n−1
,
(15)
d2n
≤ d
16.9n−1
,
n = 1, 2, . . .
and the order of convergence of the iteration (11) is τ = 9.
Proof. The proof is by induction with respect to the iteration number n.
For n = 0, from (14), we find
d1 ≤ d.d = d2 ,
d2 ≤ d81 = (d2 )8 = d16 .
53
Multipoint Algorithms Arising from...
For n = 1, we have
d3 ≤ d18 ,
d4 ≤ d144
and (15) is fulfilled.
Let (15) be fulfilled for n ≤ m.
For n = m + 1, from (14) and (15), we have
m−1
d2(m+1)−1 = d2m+1 ≤ d2m d2m−1 ≤ d2.9
.d16.9
m
d2(m+1) = d2m+2 ≤ d82m+1 < d2.9
m−1
8
m−1
= d18.9
m
= d2.9 ,
m
= d16.9
which completes the induction.
On the other hand,
1
d2n−1 = K14 |2n−1 |,
d2n
= K1 |2n |
and equation (14) can be written as
−1
|2n−1 | ≤ K1 4 d2.9
|2n |
n−1
,
n−1
≤ K1−1 d16.9
, n = 1, 2, . . . ,
and the order of convergence of iteration (11) is equal to 9.
Thus, the theorem is proved.
Remark 2. The new method (11) requires 6 function evaluations, 2 initial
approximations x−1 , x0 and has order of convergence τ = 9 and efficiency index
1
I = 9 6 ≈ 1.442.
Remark 3. The efficiency index of method (11) is better than index
I ≈ 1.412 by Newton’s procedure.
II. We consider the following modification of Euler - Chebyshev method
[25]:
xn+1 = xn −
f (xn )
f 2 (xn ) f 0 (xn ) − f 0 (xn−1 )
−
.
,
f 0 (xn ) 2f 03 (xn )
xn − xn−1
n = 0, 1, 2, . . .
(16)
54
Boryana Ignatova et al.
Here the second derivative is replaced by
f 00 (xi ) ≈
f 0 (xi ) − f 0 (xi−1 )
.
xi − xi−1
II.1. We will consider the following nonstationary iterative scheme based
on schemes (16) and (4):
x2n+1 = x2n −
f 2 (x2n ) f 0 (x2n ) − f 0 (x2n−1 )
f (x2n )
−
.
,
f 0 (x2n ) 2f 03 (x2n )
x2n − x2n−1
!
x2n+2
f (x2n+1 )
f x2n+1 − 0
f (x2n+1 )
f (x2n+1 )
f (x2n+1 )
!,
= x2n+1 − 0
− 0
.
f (x2n+1 ) f (x2n+1 )
f (x2n+1 )
f (x2n+1 ) − 2f x2n+1 − 0
f (x2n+1 )
(17)
n = 0, 1, 2, . . .
It is known that for the error i [25] is valid
3
2n+1 ∼ − c3 (ξ)22n 2n−1 ,
2
and for the procedure (4) [25]:
(18)
2n+2 ∼ c2 (ξ)[c22 (ξ) − c3 (ξ)]e42n+1 .
(19)
Let
K2
= max
n
3
|c (ξ))|, |c2 (ξ)[c22 (ξ)
2 3
o
− c3 (ξ)]| ,
3
d2n−1 = K 8 |2n−1 |,
d2n
1
= K 2 |2n |
and let d > 0, and x−1 and x0 be chosen so that the following inequalities
3
d−1 = K28 |x−1 − ξ| ≤ d < 1,
1
d0
= K22 |x0 − ξ| ≤ d < 1
hold true.
From (18) and (19), we have
55
Multipoint Algorithms Arising from...
3
8
d2n+1 = K2 |2n+1 | ≤
3
8
K2 K2 |22n ||2n−1 |
1
3
d2n+2 = K22 |2n+2 | ≤ K22 K2 42n+1 = K28
4
1
2
K2 22n
= K2 |2n−1 |
1
3
8
2
= d22n d2n−1 ,
42n+1 = d42n+1 .
(20)
Our results concerning the order of convergence generated by (17) are summarized in the following theorem.
Theorem 2.3 Assume that the initial approximations x0 , x−1 are chosen
so that d−1 ≤ d < 1 and d0 ≤ d < 1.
∞
Then for the error of the sequences {x2n }∞
n=0 and {x2n−1 }n=0 determined
by (17), we have
d2n−1 ≤ d3.9
n−1
,
(21)
d2n
≤ d
12.9n−1
,
n = 1, 2, . . .
and the order of convergence of the iteration (17) is τ = 9.
Proof. The proof is by induction with respect to the iteration number n.
For n = 0, from (20), we find
d1 ≤ d2 .d = d3 ,
d2 ≤ d41 = (d3 )4 = d12 .
For n = 1, we have
d3 ≤ d27 ,
d4 ≤ d108
and (21) is fulfilled.
Let (21) be fulfilled for n ≤ m.
For n = m + 1, from (20) and (21), we have
m−1
d2(m+1)−1 = d2m+1 ≤ d22m d2m−1 ≤ d24.9
m−1
.d3.9
m
d2(m+1) = d2m+2 ≤ d42m+1 < d3.9
which completes the induction.
On the other hand,
3
d2n−1 = K28 |2n−1 |,
1
d2n
= K22 |2n |
4
= d27.9
m−1
m
= d12.9
m
= d3.9 ,
56
Boryana Ignatova et al.
and equation (21) can be written as
−3
n−1
|2n−1 | ≤ K2 8 d3.9
−1
n−1
≤ K2 2 d12.9
|2n |
,
, n = 1, 2, . . . ,
and the order of convergence of iteration (17) is equal to 9.
Thus, the theorem is proved.
Remark 4. The method (17) requires 6 function evaluations, 2 initial
approximations x−1 , x0 and has order of convergence τ = 9 and efficiency
index
1
I = 9 6 ≈ 1.442.
II.2. We will consider the following nonstationary iterative scheme based
on schemes (16) and (10):
x2n+1 = x2n −
f 2 (x2n ) f 0 (x2n ) − f 0 (x2n−1 )
f (x2n )
−
.
,
f 0 (x2n ) 2f 03 (x2n )
x2n − x2n−1
y2n+1 = x2n+1 −
f (x2n+1 )
,
f 0 (x2n+1 )
z2n+1 = y2n+1 −
f (y2n+1 )
f (x2n+1 )
.
,
0
f (x2n+1 ) f (x2n+1 ) − 2f (y2n+1 )

x2n+2
f (z2n+1 )
f (z2n+1 )
= z2n+1 − 1 +
+
f (x2n+1 )
f (x2n+1
!2 
.
f [x2n+1 , y2n+1 ]f (z2n+1 )
,
f [x2n+1 , z2n+1 ]f [y2n+1 , z2n+1 ]
(22)
n = 0, 1, 2, . . .
It is known that for the error 2n+2 = x2n+2 − ξ [18] is valid
2n+2 ∼ A(ξ)e82n+1 ,
(23)
We will use again the fact that for method (16) is satisfied
3
2n+1 ∼ − c3 (ξ)22n 2n−1 ,
2
(24)
Let
K3
= max
n
3
d2n−1 = K316 |2n−1 |,
1
d2n
o
3
|c (ξ)|, |A(ξ)|
2 3
= K32 |2n |
,
57
Multipoint Algorithms Arising from...
and let d > 0, and x−1 and x0 be chosen so that the following inequalities
3
d−1 = K316 |x−1 − ξ| ≤ d < 1,
1
= K32 |x0 − ξ| ≤ d < 1
d0
hold true.
From (23) and (24), we have
3
16
d2n+1 = K3 |2n+1 | ≤
1
2
d2n+2 = K3 |2n+2 | ≤
3
16
K3 K3 22n |2n−1 |
1
2
K3 K3 82n+1
3
16
1
2
= K3 |2n−1 | K3 2n
3
16
8
= K3
2
= d22n d2n−1 ,
82n+1 = d82n+1 .
(25)
Our results concerning the order of convergence generated by (22) are summarized in the following theorem.
Theorem 2.4 Assume that the initial approximations x0 , x−1 are chosen
so that d−1 ≤ d < 1 and d0 ≤ d < 1.
∞
Then for the error of the sequences {x2n }∞
n=0 and {x2n−1 }n=0 determined
by (22), we have
n−1
d2n−1 ≤ d3.17
,
(26)
d2n
≤ d
24.17n−1
,
n = 1, 2, . . .
and the order of convergence of the iteration (22) is τ = 17.
Proof. The proof is by induction with respect to the iteration number n.
For n = 0, from (25), we find
d1 ≤ d2 .d = d3 ,
d2 ≤ d81 = (d3 )8 = d24 .
For n = 1, we have
d3 ≤ d51 ,
d4 ≤ d408
and (26) is fulfilled.
Let (26) be fulfilled for n ≤ m.
For n = m + 1, from (25) and (26), we have
58
Boryana Ignatova et al.
m−1
d2(m+1)−1 = d2m+1 ≤ d22m d2m−1 ≤ d48.17
m−1
.d3.17
m
d2(m+1) = d2m+2 ≤ d82m+1 < d3.17
8
m−1
= d51.17
m
= d3.17 ,
m
= d24.17
which completes the induction.
On the other hand,
3
d2n−1 = K316 |2n−1 |,
1
= K32 |2n |
d2n
and equation (26) can be written as
3
− 16
3.17n−1
|2n−1 | ≤ K3
|2n |
−1
d
n−1
≤ K3 2 d24.17
,
, n = 1, 2, . . . ,
and the order of convergence of iteration (22) is equal to 17.
Thus, the theorem is proved.
Remark 5. The method (22) requires 7 function evaluations, 2 initial
approximations x−1 , x0 and has order of convergence τ = 17 and efficiency
index
1
I = 17 7 ≈ 1.4989.
III. The following modification of Euler - Chebyshev method, where
the second derivative is replaced by its approximation by differentiating a twopoint Hermite interpolation formula for f (x), the two point being xi and xi−1
is given by
xn+1 = xn −
f (xn )
f 2 (xn ) (2f 0 (xn ) + f 0 (xn−1 ) − 3f (xn , xn−1 ))
−
,
f 0 (xn )
f 03 (xn )(xn − xn−1 )
(27)
n = 0, 1, 2, . . .
III.1. We will consider the following nonstationary iterative scheme based
on schemes (27) and (4):
59
Multipoint Algorithms Arising from...
x2n+1 = x2n −
f (x2n )
f 2 (x2n ) (2f 0 (x2n ) + f 0 (x2n−1 ) − 3f (x2n , x2n−1 ))
−
,
f 0 (x2n )
f 03 (x2n )(x2n − x2n−1 )
!
x2n+2
f (x2n+1 )
f x2n+1 − 0
f (x2n+1 )
f (x2n+1 )
f (x2n+1 )
!,
− 0
.
= x2n+1 − 0
f (x2n+1 ) f (x2n+1 )
f (x2n+1 )
f (x2n+1 ) − 2f x2n+1 − 0
f (x2n+1 )
(28)
n = 0, 1, 2, . . .
It is known that for the error i [17] is valid
2n+1 ∼ B(ξ)22n 22n−1 ,
(29)
and for the procedure (4) [25]:
2n+2 ∼ c2 (ξ)[c22 (ξ) − c3 (ξ)]e42n+1 .
(30)
Let
K4
= max {|c2 (ξ)[c22 (ξ) − c3 (ξ)]|, |B(ξ)|} ,
1
d2n−1 = K43 |2n−1 |,
1
d2n
= K43 |2n |
and let d > 0, and x−1 and x0 be chosen so that the following inequalities
1
d−1 = K43 |x−1 − ξ| ≤ d < 1,
1
= K43 |x0 − ξ| ≤ d < 1
d0
hold true.
From (29) and (30), we have
1
1
1
d2n+1 = K43 |2n+1 | ≤ K43 K4 22n 22n−1 = K43 2n−1
1
3
d2n+2 = K4 |2n+2 | ≤
1
3
K4 K4 42n+1
1
3
= K4
4
2 1
K43 2n
2
= d22n d22n−1 ,
42n+1 = d42n+1 .
(31)
Our results concerning the order of convergence generated by (28) are summarized in the following theorem.
60
Boryana Ignatova et al.
Theorem 2.5 Assume that the initial approximations x0 , x−1 are chosen
so that d−1 ≤ d < 1 and d0 ≤ d < 1.
∞
Then for the error of the sequences {x2n }∞
n=0 and {x2n−1 }n=0 determined
by (28), we have
n−1
d2n−1 ≤ d4.10
,
(32)
d2n
≤ d
16.10n−1
,
n = 1, 2, . . .
and the order of convergence of the iteration (28) is τ = 10.
Proof. The proof is by induction with respect to the iteration number n.
For n = 0, from (31), we find
d1 ≤ d2 .d2 = d4 ,
d2 ≤ d41 = (d4 )4 = d16 .
For n = 1, we have
d3 ≤ d40 ,
d4 ≤ d160
and (32) is fulfilled.
Let (32) be fulfilled for n ≤ m.
For n = m + 1, from (31) and (32), we have
m−1
d2(m+1)−1 = d2m+1 ≤ d22m d22m−1 ≤ d32.10
m−1
.d8.10
m
d2(m+1) = d2m+2 ≤ d42m+1 < d4.10
4
1
d2n−1 = K43 |2n−1 |,
1
= K43 |2n |
and equation (32) can be written as
−1
n−1
|2n−1 | ≤ K4 3 d4.10
|2n |
−1
,
n−1
≤ K4 3 d16.10
m
= d16.10
which completes the induction.
On the other hand,
d2n
m−1
= d40.10
, n = 1, 2, . . . ,
m
= d4.10 ,
61
Multipoint Algorithms Arising from...
and the order of convergence of iteration (28) is equal to 10.
Thus, the theorem is proved.
Remark 6. The method (28) requires 7 function evaluations, 2 initial
approximations x−1 , x0 and has order of convergence τ = 10 and efficiency
index
1
I = 10 7 ≈ 1.389.
III.2. We will consider the following nonstationary iterative scheme based
on schemes (10) and (27):
x2n+1
f 2 (x2n ) (2f 0 (x2n ) + f 0 (x2n−1 ) − 3f (x2n , x2n−1 ))
f (x2n )
−
,
= x2n − 0
f (x2n )
f 03 (x2n )(x2n − x2n−1 )
y2n+1 = x2n+1 −
f (x2n+1 )
,
f 0 (x2n+1 )
z2n+1 = y2n+1 −
f (y2n+1 )
f (x2n+1 )
.
,
0
f (x2n+1 ) f (x2n+1 ) − 2f (y2n+1 )

x2n+2
f (z2n+1 )
f (z2n+1 )
= z2n+1 − 1 +
+
f (x2n+1 )
f (x2n+1
!2 
.
f [x2n+1 , y2n+1 ]f (z2n+1 )
,
f [x2n+1 , z2n+1 ]f [y2n+1 , z2n+1 ]
(33)
n = 0, 1, 2, . . .
It is known that for the error 2n+2 = x2n+2 − ξ [18] is valid:
2n+2 ∼ A(ξ)e82n+1 ,
(34)
We will use again the fact that for the method (27) is satisfied
2n+1 ∼ B(ξ)22n 22n−1 .
(35)
Let
K5
= max {|B(ξ)|, |A(ξ)|} ,
3
d2n−1 = K517 |2n−1 |,
7
d2n
= K517 |2n |
and let d > 0, and x−1 and x0 be chosen so that the following inequalities
3
d−1 = K517 |x−1 − ξ| ≤ d < 1,
7
d0
= K517 |x0 − ξ| ≤ d < 1
62
Boryana Ignatova et al.
hold true.
From (34) and (35), we have
3
3
3
d2n+1 = K517 |2n+1 | ≤ K517 K5 22n 22n−1 = K517 2n−1
7
7
3
d2n+2 = K517 |2n+2 | ≤ K517 K5 82n+1 = K517
8
2 2
7
K517 2n
= d22n d22n−1 ,
82n+1 = d82n+1 .
(36)
Our results concerning the order of convergence generated by (33) are summarized in the following theorem.
Theorem 2.6 Assume that the initial approximations x0 , x−1 are chosen
so that d−1 ≤ d < 1 and d0 ≤ d < 1.
∞
Then for the error of the sequences {x2n }∞
n=0 and {x2n−1 }n=0 determined
by (33), we have
n−1
d2n−1 ≤ d4.18
,
(37)
d2n
≤ d
32.18n−1
,
n = 1, 2, . . .
and the order of convergence of the iteration (33) is τ = 18.
Proof. The proof is by induction with respect to the iteration number n.
For n = 0, from (36), we find
d1 ≤ d2 .d2 = d4 ,
d2 ≤ d81 = (d4 )8 = d32 .
For n = 1, we have
d3 ≤ d72 ,
d4 ≤ d576
and (37) is fulfilled.
Let (37) be fulfilled for n ≤ m.
For n = m + 1, from (36) and (37), we have
m−1
d2(m+1)−1 = d2m+1 ≤ d22m d22m−1 ≤ d64.18
m−1
.d8.18
m
d2(m+1) = d2m+2 ≤ d82m+1 < d4.18
which completes the induction.
8
m−1
= d72.18
m
= d32.18
m
= d4.18 ,
63
Multipoint Algorithms Arising from...
On the other hand,
3
d2n−1 = K517 |2n−1 |,
7
= K517 |2n |
d2n
and equation (37) can be written as
3
− 17
4.18n−1
|2n−1 | ≤ K5
|2n |
d
,
7
− 17
32.18n−1
≤ K5
d
, n = 1, 2, . . . ,
and the order of convergence of iteration (33) is equal to 18.
Thus, the theorem is proved.
Remark 7. The method (33) requires 8 function evaluations, 2 initial
approximations x−1 , x0 and has order of convergence τ = 18 and efficiency
index
1
I = 18 8 ≈ 1.435.
From the Table 1 which is given below, the user of the most common practice multipoint iterative methods using several initial approximations for numerical solution of nonlinear equations can be oriented to the self-accelerating
with the help of optimal in the sense of Kung-Traub algorithms from order 4
and 8 in terms of convergence speed and efficiency index.
Table 1
method
(5)
(11)
(17)
(22)
(28)
(33)
order of convergence τ
5
9
9
17
10
18
efficiency index I
1.3797
1.442
1.442
1.4989
1.389
1.435
We will pose the following problem:
Problem. Let us construct an iteration procedure (with memory) with
1
order of convergence τ = 33 and efficiency index - better than 2 2 ≈ 1.414 of
Newton’s method using:
a) a system of two initial approximations x−1 and x0 ;
b) information about f and f 0 ;
In [12] Li, Mu, Ma and Wang presented a modification of Newton’s method
with higher-order of convergence.
64
Boryana Ignatova et al.
The modification of Newton’s method is based on known King’s fourth order method. The new method requires three-step per iteration.
Analysis of convergence demonstrates that the order of convergence is 16.
If the initial point x0 is sufficiently close to simple root x∗ , then the method
[12] defined by
yn = x n −
f (xn )
,
f 0 (xn )
zn = yn −
2f (xn ) − f (yn ) f (yn )
.
,
2f (xn ) − 5f (yn ) f 0 (xn )
(38)
!
xn+1 = zn −
f (zn )
−
f 0 (zn )
!
f (zn )
f (zn )
2f (zn ) − f zn − 0
0
f (zn )
f (zn )
!,
.
0
f (zn )
f (zn )
2f (zn ) − 5f zn − 0
f (zn )
f zn −
n = 0, 1, 2, . . .
has sixteenth - order of convergence.
Remark 8. This method requires four evaluations of the function, namely,
f (zn )
f (xn ), f (yn ), f (zn ), f zn − 0
f (zn )
!
and two evaluations of first derivatives f 0 (xn ), f 0 (zn ) and is not optimal in the
sense of Kung - Traub.
Here we give a methodological construction of nonstationary algorithms
with a raised speed of convergence.
For solving this task it is appropriate to use the following iterative nonsta-
65
Multipoint Algorithms Arising from...
tionary algorithm for solving the nonlinear equation f (x) = 0:
x2n+1 = x2n −
f (x2n )
f 2 (x2n ) f 0 (x2n ) − f 0 (x2n−1 )
−
.
,
f 0 (x2n ) 2f 03 (x2n )
x2n − x2n−1
y2n+1 = x2n+1 −
f (x2n+1 )
,
f 0 (x2n+1 )
z2n+1 = y2n+1 −
2f (x2n+1 ) − f (y2n+1 ) f (y2n+1 )
.
,
2f (x2n+1 ) − 5f (y2n+1 ) f 0 (x2n+1 )
!
x2n+2
f (z2n+1 )
f z2n+1 − 0
f (z2n+1 )
f (z2n+1 )
= z2n+1 − 0
−
×
0
f (z2n+1 )
f (z2n+1 )
(39)
!
f (z2n+1 )
2f (z2n+1 ) − f z2n+1 − 0
f (z2n+1 )
!,
×
f (z2n+1 )
2f (z2n+1 ) − 5f z2n+1 − 0
f (z2n+1 )
n = 0, 1, 2, . . . .
Here {x2n+1 } is generated by (38), {x2n+2 } based on the algorithm (16).
It is known that for the error i [25] is valid
3
2n+1 ∼ − c3 (ξ)22n 2n−1 ,
2
and for the procedure (38) [12]:
(40)
2n+2 ∼ −c2 (ξ)5 c3 (ξ)5 e16
2n+1 ,
(41)
where ∼ denotes the asymptotical equation when n → ∞.
Let
K6
= max
n
o
3
|c (ξ)|, |c2 (ξ)5 c3 (ξ)5 |
2 3
,
3
d2n−1 = K632 |2n−1 |,
1
d2n
= K62 |2n |
and let d > 0, and x−1 and x0 be chosen so that the following inequalities
3
d−1 = K632 |x−1 − ξ| ≤ d < 1,
1
d0
= K62 |x0 − ξ| ≤ d < 1
66
Boryana Ignatova et al.
hold true.
From (40) and (41), we have
3
3
3
d2n+1 = K632 |2n+1 | ≤ K632 K6 22n |2n−1 | = K632 |2n−1 |
1
2
d2n+2 = K6 |2n+2 | ≤
1
2
K6 K6 16
2n+1
16
3
32
= K6 2n+1
2
1
K62 2n
= d22n d2n−1 ,
= d16
2n+1 .
(42)
Our results concerning the order of convergence generated by (39) are summarized in the following theorem.
Theorem 2.7 Assume that the initial approximations x0 , x−1 are chosen
so that d−1 ≤ d < 1 and d0 ≤ d < 1.
∞
Then for the error of the sequences {x2n+1 }∞
n=0 and {x2n+2 }n=0 determined
by (39), we have
n−1
d2n−1 ≤ d3.33
,
(43)
d2n
≤ d
48.33n−1
,
n = 1, 2, . . .
and the order of convergence of the iteration (39) is τ = 33.
Proof. The proof is by induction with respect to the iteration number n.
For n = 0, from (42), we find
d1 ≤ d2 .d = d3 ,
3 16
d2 ≤ d16
= d48 .
1 ≤ (d )
For n = 1, we have
2
d3 ≤ d22 d1 ≤ (d48 ) d3 = d99 ,
99
d4 ≤ d16
3 ≤ (d )
16
= d1584
and (43) is fulfilled.
Let (43) be fulfilled for n ≤ m.
For n = m + 1, from (42) and (43), we have
m−1 +3.33m−1
d2(m+1)−1 = d2m+1 ≤ d22m d2m−1 ≤ d96.33
m
3.33
d2(m+1) = d2m+2 ≤ d16
2m+1 < d
16
m−1
= d3.33.33
m
= d48.33
m
= d3.33 ,
67
Multipoint Algorithms Arising from...
which completes the induction.
On the other hand,
3
d2n−1 = K632 |2n−1 |,
1
= K62 |2n |
d2n
and equation (43) can be written as
3
− 32
3.33n−1
|2n−1 | ≤ K6
|2n |
d
−1
n−1
≤ K6 2 d48.33
,
, n = 1, 2, . . . ,
and the order of convergence of iteration (39) is equal to 33.
Thus, the theorem is proved.
Remark 9. The method (39) requires 9 function evaluations, 2 initial
approximations x−1 , x0 and has order of convergence τ = 33 and efficiency
index
1
I = 33 9 ≈ 1.4747
1
which is better than 2 2 ≈ 1.414 of Newton’s method.
3
Concluding Remarks
As we have previously noted, an iterative procedure (38) with order of convergence τ = 16 is not optimal in the sense of Kung - Traub, because it uses six
calculations of functions.
Let us assume for a moment that iteration algorithm (B) with order of
convergence τ = 16 using five functional calculations i.e. optimal in the sense
of Kung - Traub was found.
To examine the issue related to self-accelerating and the corresponding
efficiency index of the base method - the modification of Euler - Chebishev
method - (16), with already familiar ”put in” of the optimal method (B).
As a result, we get nonstationary algorithm, which use two initial approximations x0 and x−1 . For brevity we denote it (16) − (B).
It is not difficult to comply, that the new iterative scheme (16) − (B)
will have order of convergence τ = 33, and it consumes eight calculations of
functions and it has an efficiency index
1
I = 33 8 ≈ 1.5481.
68
Boryana Ignatova et al.
It is a matter of time, having in mind the massive research in the area
of numerical methods over the last five years, the design of algorithms with
optimal order of convergence τ = 32 and τ = 64 using six respectively seven
calculations of functions.
Let us denote by (C) the algorithm with this order of convergence τ = 32.
Naturally arises the task of testing the combined procedure (16) − (C).
The following model theorem will be valid
Theorem A. With the same symbols in this paper and imposed requirements for initial approximations x0 , x−1 , the order of convergence of the model
iteration (16) − (C) is satisfied
n−1
d2n−1 ≤ d3.65
,
(44)
d2n
≤ d
96.65n−1
,
n = 1, 2, . . .
Remark 10. The method (16) − (C) requires 9 function evaluations,
2 initial approximations x−1 , x0 and has order of convergence τ = 65 and
efficiency index
1
I = 65 9 ≈ 1.5901.
Remark 11. The reader may account how self-accelerating are to the order
of convergence and efficiency index methods of type (2) − (C) and (27) − (C)
using two initial approximations.
Denote now with (D) this optimal in the sense of Kung - Traub algorithm
with order of convergence τ = 64 using 7 calculations of functions.
We will examine the combined procedure (16) − D).
The following model theorem is valid
Theorem 3.1 With the same symbols in this paper and imposed requirements for the initial approximations x0 , x−1 , the order of convergence of the
model iteration (16) − (D) is satisfied
n−1
d2n−1 ≤ d3.129
,
(45)
d2n
≤ d
192.129n−1
,
n = 1, 2, . . .
Remark 12. The method (16) − (D) requires 10 function evaluations,
2 initial approximations x−1 , x0 and has order of convergence τ = 129 and
efficiency index
1
I = 129 10 ≈ 1.6257.
69
Multipoint Algorithms Arising from...
We are now able to offer the following Table 2 for possible self-accelerating,
as the basis on multipoint method (16) with the “put in” of newly found optimal in the sense of Kung-Traub algorithms from order 16, 32 and 64 in terms
of convergence speed and efficiency index.
Table 2
method
(16)-(B)
(16)-(C)
(16)-(D)
order of convergence τ
33
65
129
efficiency index I
1.5481
1.5901
1.6257
Remark 13. Let us denote by τ1 the order of convergence of the corresponding optimal algorithm of order 4, 8, 16, 32, 64, and with τ2 - the order of
convergence of the basic multipoint iteration algorithm, as an example (16),
generated by the optimal in the sense of Kung - Traub algorithm.
A detailed analysis (see Theorem 2.3, Theorem 2.4, Theorem A, Theorem
B) gives us reason to conclude that
τ2 = 2τ1 + 1.
Remark 14. Of course, based on the analysis of the data in Table 1
and Table 2, the user of such multipoint iterative schemes should consider
for himself ”what is the price that he is willing to pay” to achieve speed and
efficiency of the used nonstationary algorithm.
Remark 15. The results obtained in this article can be successfully used
to refine the self-accelerating of multipoint iterations using three initial approximations x−2 , x−1 , x0 .
IV. Consider the following finite-difference analogue of Halley method
xn+1 = xn −
f (xn )
,
f [xn , xn−1 ] + f [xn , xn−1 , xn−2 ](xn − xn−1 )
(46)
n = 0, 1, 2, . . .
which requires three initial approximations x−2 , x−1 , x0 .
IV.1. We will explore the issue of self acceleration of this procedure, in combination with optimal algorithm in the sense of Kung - Traub, as an example
with order of convergence 4 - (4):
70
Boryana Ignatova et al.
x2n+1 = x2n −
f (x2n )
,
f [x2n , x2n−1 ] + f [x2n , x2n−1 , x2n−2 ](x2n − x2n−1 )
!
x2n+2
f (x2n+1 )
f x2n+1 − 0
f (x2n+1 )
f (x2n+1 )
f (x2n+1 )
!,
= x2n+1 − 0
− 0
.
f (x2n+1 ) f (x2n+1 )
f (x2n+1 )
f (x2n+1 ) − 2f x2n+1 − 0
f (x2n+1 )
(47)
n = 0, 1, 2, . . .
It is known that for the error i = xi − ξ, i = −2, −1, 0, 1, 2 . . . ; [25] is
valid
2n+1 ∼ L(ξ)2n 2n−1 2n−2 ,
(48)
2n+2 ∼ M (ξ)42n+1 .
(49)
Let
K7
= max {|L(ξ))|, |M (ξ)|} ,
3
d2n−1 = K78 |2n−1 |,
1
d2n
= K72 |2n |,
1
d2n−2 = K72 |2n−2 |,
and let d > 0, and x−2 , x−1 and x0 be chosen so that the following inequalities
1
d−2 = K72 |x−2 − ξ| ≤ d < 1,
3
d−1 = K78 |x−1 − ξ| ≤ d < 1,
1
d0
= K72 |x0 − ξ| ≤ d < 1
hold true.
From (48) and (49), we have
71
Multipoint Algorithms Arising from...
3
3
3
1
1
d2n+1 = K78 |2n+1 | ≤ K78 K7 |2n ||2n−1 ||2n−2 | = K78 |2n−1 |K72 |2n |K72 |2n−2 | =
= d2n d2n−1 d2n−2 ,
1
1
3
d2n+2 = K72 |2n+2 | ≤ K72 K7 42n+1 = K78 2n+1
4
= d42n+1 .
(50)
Evidently, from (50), we find
d1 ≤ d3 , d2 ≤ d12 , d3 ≤ d16 , d4 ≤ d64 ,
d5 ≤ d92 , d6 ≤ d368 , d7 ≤ d524 , d8 ≤ d2096 ,
d9 ≤ d2988 , d10 ≤ d11952 .
Our results concerning the order of convergence generated by (47) are summarized in the following theorem.
Theorem 3.2 Assume that the initial approximations x0 , x−1 , x−2 are chosen so that d−2 ≤ d, d−1 ≤ d < 1 and d0 ≤ d < 1.
∞
Then for the error of the sequences {x2n+1 }∞
n=0 and {x2n+2 }n=0 determined
by (47), we have
d2n−1 ≤ dτ2n−1 ,
(51)
d2n
≤ dτ2n ,
where
τm+6 = 4τm+4 + 9τm+2 + 4τm ,
m = 0, 1, 2, . . .
and the order of convergence of the iteration (47) is
τ=
.
5+
√
41
.
2
(52)
72
Boryana Ignatova et al.
Proof. It is well known that for the recursion:
γi+1 =
n
X
Aj γi−j+1 , i = n − 1, n − 2, . . . ,
j=1
(for any initial conditions) corresponds to the characteristic polynomial:
n
ρ =
n
X
Aj ρn−j .
j=1
In our case, for the recursion
τm+6 = 4τm+4 + 9τm+2 + 4τm ,
characteristic polynomial is of the type
ρ3 − 4ρ2 − 9ρ − 4 = 0.
(53)
Equation (53) has the roots:
√
√
5 − 41
5 + 41
, ρ2 =
, ρ3 = −1.
ρ1 =
2
2
From the general iterative theory [25], (see, also [5]) it follows that the
order of convergence of the iteration procedure, defined by (47) is given by the
only real root of equation (53) with magnitude greater than 1.
On the other hand,
−3
|2n+1 | ≤ K7 8 d2n+1 ,
−1
|2n+2 | ≤ K7 2 d2n+2 ,
and consequently we can conclude that the order of convergence of iteration
(47) is
√
5 + 41
τ=
≈ 5.70156...
2
Thus, the theorem is proven.
Remark 16. Of course precise analysis can be made following the cited
monographs [25], [5].
Suffice it to seek representation for the values τi as a linear combination of
the roots of characteristic equation ρ1 , ρ2 and ρ3 .
As an example,
τ2n+1 = c1 ρn1 + c2 ρn2 + c3 ρn3
73
Multipoint Algorithms Arising from...
with condition
τ1 = 3, τ3 = 16, τ5 = 92.
The solution of the system
c1 + c2 + c3 = 3
√
√
5 + 41
5 − 41
c1
+ c2
− c3 = 16
2
2
√ !2
√ !2
5 + 41
5 − 41
c1
+ c2
+ c3 = 92
2
2
is the following:
√
1
(123 + 17 41)
82
√
1
c2 = (123 − 17 41)
82
c1 =
c3 = 0.
This lead to the following inequality
√
1
d2n ≤ d 82 (41+19
√
1
n
41)ρn
1 + 82 (41−19 41)ρ2
,
(54)
The proof is by induction with respect to the iteration number n.
Let (54) be fulfilled for n ≤ m.
For n = m + 1, we have
√
4
d2(m+1) = d2m+2 ≤ d42m+1 ≤ d 82 (123+17
√
4
m
41)ρm
1 + 82 (123−17 41)ρ2
Using the equalities
√
√
√
1
1
123 ± 17 41 = (41 ± 19 41). (5 ± 41)
4
2
we have
√
1
d2(m+1) ≤ d 82 (41+19
√
1
41)ρm+1
(41−19 41)ρm+1
+ 82
1
2
which completes the induction.
Evidently, −1 < ρ2 < 0, so that, asymptotically,
√
1
d 82 (41−19
41)ρm
2
≈1
and
−1
1
√
|2n | ≈ K7 2 d 82 (41+19
41)ρn
1
.
.
74
Boryana Ignatova et al.
Remark 17. The method (47) requires 6 function evaluations, 3 initial
approximations x−2 , x−1 x0 and has order of convergence τ = 5.70156 and
efficiency index
1
I = 5.70156 6 ≈ 1.3365.
IV.2. We will consider the following nonstationary iterative scheme based
on the scheme (46) in combination with an optimal algorithm in the sense of
Kung - Traub with order of convergence 8 - (10) [18]:
x2n+1 = x2n −
f (x2n )
,
f [x2n , x2n−1 ] + f [x2n , x2n−1 , x2n−2 ](x2n − x2n−1 )
y2n+1 = x2n+1 −
f (x2n+1 )
,
f 0 (x2n+1 )
z2n+1 = y2n+1 −
f (y2n+1 )
f (x2n+1 )
.
,
0
f (x2n+1 ) f (x2n+1 ) − 2f (y2n+1 )

x2n+2 = z2n+1 − 1 +
f (z2n+1 )
f (z2n+1 )
+
f (x2n+1 )
f (x2n+1
!2 
.
f [x2n+1 , y2n+1 ]f (z2n+1 )
,
f [x2n+1 , z2n+1 ]f [y2n+1 , z2n+1 ]
(55)
n = 0, 1, 2, . . .
It is known that for the error i = xi − ξ, i = −2, −1, 0, 1, 2 . . . ; [25], [18]
is valid
2n+1 ∼ L(ξ)2n 2n−1 2n−2 ,
(56)
2n+2 ∼ A(ξ)82n+1 .
(57)
Let
K8
= max {|L(ξ))|, |A(ξ)|} ,
3
d2n−1 = K816 |2n−1 |,
1
d2n
= K82 |2n |,
1
d2n−2 = K82 |2n−2 |,
75
Multipoint Algorithms Arising from...
and let d > 0, and x−2 , x−1 and x0 be chosen so that the following inequalities
1
d−2 = K82 |x−2 − ξ| ≤ d < 1,
3
d−1 = K816 |x−1 − ξ| ≤ d < 1,
1
= K82 |x0 − ξ| ≤ d < 1
d0
hold true.
From (56) and (57), we have
3
3
3
1
1
d2n+1 = K816 |2n+1 | ≤ K816 K8 |2n ||2n−1 ||2n−2 | = K816 |2n−1 |K82 |2n |K82 |2n−2 | =
= d2n d2n−1 d2n−2 ,
1
2
d2n+2 = K8 |2n+2 | ≤
1
2
K8 K8 82n+1
3
16
= K8 2n+1
8
= d82n+1 .
(58)
Evidently, from (58), we find
d1 ≤ d3 , d2 ≤ d24 , d3 ≤ d28 , d4 ≤ d224 ,
d5 ≤ d276 , d6 ≤ d2208 , d7 ≤ d2708 , d8 ≤ d21664 ,
d9 ≤ d26580 , d10 ≤ d212640 .
Our results concerning the order of convergence generated by (55) are summarized in the following theorem.
Theorem 3.3 Assume that the initial approximations x0 , x−1 , x−2 are chosen so that d−2 ≤ d, d−1 ≤ d < 1 and d0 ≤ d < 1.
∞
Then for the error of the sequences {x2n+1 }∞
n=0 and {x2n+2 }n=0 determined
by (55), we have
d2n−1 ≤ dτ2n−1 ,
(59)
d2n
≤ dτ2n ,
where
τm+6 = 8τm+4 + 17τm+2 + 8τm ,
(60)
76
Boryana Ignatova et al.
m = 0, 1, 2, . . .
and the order of convergence of the iteration (55) is
√
9 + 113
τ=
.
2
.
Proof. The detailed proof of this theorem will not be given because the
reasoning follows the statement of the proof of Theorem 3.2.
We note only that in this case the recursion
τm+6 = 8τm+4 + 17τm+2 + 8τm ,
satisfies the characteristic polynomial
ρ3 − 8ρ2 − 17ρ − 8 = 0.
(61)
Equation (61) has the roots:
√
√
9 + 113
9 − 113
ρ1 =
, ρ2 =
, ρ3 = −1.
2
2
From the general iterative theory [25], (see, also [5]) it follows that the
order of convergence of the iteration procedure, defined by (55) is given by the
only positive root of equation (61).
On the other hand,
3
− 16
|2n+1 | ≤ K8
d2n+1 ,
−1
|2n+2 | ≤ K8 2 d2n+2 ,
and consequently we can conclude that the order of convergence of iteration
(55) is
√
9 + 113
τ=
≈ 5.78702...
2
Thus, the theorem is proved.
Remark 18. The method (55) requires 7 function evaluations, 3 initial
approximations x−2 , x−1 x0 and has order of convergence τ = 5.78702 and
efficiency index
1
I = 5.78702 7 ≈ 1.285.
Remark 19. The received during last year numerical methods which have
high index of efficiency by F. Soleymani and his coauthors [19], [21], [22], [23]
Multipoint Algorithms Arising from...
77
can be successfully used for generation of new iterative methods for solving
nonlinear equations using several initial approximations by proposed in our
paper techniques.
Acknowledgements
This paper is partly supported by project NI11–FMI–004 of Department for
Scientific Research, Paisii Hilendarski University of Plovdiv.
References
[1] W. Bi, H. Ren and Q. Wu, Three-step iterative methods with eight-order
convergence for solving nonlinear equations, J. Comput. Appl. Math.,
225(2009), 105-112.
[2] W. Bi, Q. Wu and H. Ren, A new family of eight-order iterative methods
for solving nonlinear equations, Appl. Math. Comput., 214(2009), 236-245.
[3] C. Chun and B. Neta, Certain improvements of Newton’s method with
fourth-order convergence, Appl. Math. Comput., 215(2009), 821-824.
[4] Y.H. Geum and Y.I. Lim, A multiparameter family of three-step eightorder iterative methods locating a simple root, Appl. Math. Comput.,
215(2010), 3375-3382.
[5] A. Householder, The Numerical Treatment of a Single Nonlinear Equations, McGraw-Hill Book Company, (1970).
[6] V. Hristov, A. Iliev and N. Kyurkchiev, A note on the convergence of
nonstationary finite-difference analogues, Comp. Math. and Math. Phys.,
45(2005), 194-201.
[7] A. Iliev and N. Kyurkchiev, Nontrivial Methods in Numerical Analysis:
Selected Topics in Numerical Analysis, LAP LAMBERT Academic Publishing, (2010).
[8] S. Khattri and S. Abbasbandy, Optimal Fourth-Order Family of Iterative
Methods, Mathematicki Vesnik, In Press, (2010).
[9] J. Kou, X. Wang and S. Sun, Some new root-finding methods with eightorder convergence, Bull. Math. Soc. Sci. Math. Roumanie, 53(2010), 133143.
[10] H. Kung and J. Traub, Optimal order of one point and multipoint iteration, ACM, 21(1974), 643-651.
78
Boryana Ignatova et al.
[11] N. Kyurkchiev and A. Iliev, A note on the constructing of nonstationary
methods for solving nonlinear equations with raised speed of convergence,
Serdica J. Computing, 3(2009), 47-74.
[12] X. Li, C. Mu, J. Ma and C. Wang, Sixteenth-order method for nonlinear
equations, Appl. Math. Comput., 215(2010), 3754-3758.
[13] X. Liangzang and M. Xiangjiand, Convergence of an iteration method
without derivative, Numer. Math. J. Chinese Univ. Engl. Ser., 11(2002),
113-120.
[14] A.M. Ostrowski, Solution of Equations and System of Equations, Academic Press, New York, (1960).
[15] M. Petkovic, On optimal multipoint methods for solving nonlinear equations, Novi Sad J. Math., 39(2009), 123-130.
[16] M. Petkovic and L. Petkovic, Families of optimal multipoint methods
for solving nonlinear equations: A Survey, Appl. Anal. Discrete Math.,
4(2010), 1-22.
[17] A. Ralston, A First Course in Numerical Analysis, McGraw-Hill, New
York, (1965).
[18] J.R. Sharma and R. Sharma, A new family of modified Ostrowski’s methods with accelerated eight order convergence, Numer. Algor., 54(2010),
445-458.
[19] F. Soleimani and F. Soleymani, Computing simple roots by a sixth-order
iterative method, Int. J. of Pure and Appl. Math., 69(2011), 41-48.
[20] F. Soleymani, Regarding the Accuracy of Optimal Eight-Order Methods,
Math. and Comput. Modelling, In Press, (2010).
[21] P. Sargolzaei and F. Soleymani, Accurate fourteenth-order methods for solving nonlinear equations, J. Numerical Algorithms, (2011),
DOI:10.1007/s11075-011-9467-4.
[22] F. Soleymani and B. Mousavi, A novel computational technique for finding
simple roots of nonlinear equations, Int. J. of Math. Analysis, 5(2011),
1813-1819.
[23] F. Soleymani and M. Sharifi, On a general efficient class of four-step
root-finding methods, Int. J. of Math. and Comp. in Simulation, 5(2011),
181-189.
Multipoint Algorithms Arising from...
79
[24] R. Thurkal and M. Petkovic, A family of three-point methods of optimal
order for solving nonlinear equations, J. Comput. Appl. Math., 233(2010),
2278-2284.
[25] J. Traub, Iterative Methods for Solution of Equations, Prentice-Hall, Englewood Cliffs, N. J., (1964).
[26] X. Wang and L. Liu, New eight-order iterative methods for solving nonlinear equations, J. Comput. Appl. Math., 234(2010), 1611-1620.
[27] W. Xinghua, Z. Shiming and H. Danfu, Convergence on Euler’s series,
the iteration of Euler’s and Halley’s families, Acta Math. Sinica, 33(1990),
721-738.
Download